-
Notifications
You must be signed in to change notification settings - Fork 92
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Feature/inc cache propagation #1220
base: master
Are you sure you want to change the base?
Conversation
df97144
to
152993b
Compare
152993b
to
17031c9
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm in for merging this soon and not letting this PR stale, however, it adds non trivial changes, including in the main codebase. I think it really needs more documentation and comments, if we want to be able to understand what's going on a few months from now.
Beside, I wonder if we might also update Thunk::saturate to work "in-place", now that we revert everything in advance anyway. But that's not very important for this PR, I'm just thinking out loud.
fn smart_clone(&mut self, v: Vec<CacheIndex>) -> HashMap<CacheIndex, CacheIndex>; | ||
|
||
fn propagate_dirty(&mut self, indices: Vec<CacheIndex>); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
We need to document those functions, even just a draft, waiting for more. Those are the cornerstones of incremental evaluation, and are now used inside merge.
@@ -370,4 +314,67 @@ impl Cache for IncCache { | |||
RichTerm::from(Term::App(partial_app, arg)) | |||
}) | |||
} | |||
|
|||
fn smart_clone(&mut self, v: Vec<CacheIndex>) -> HashMap<CacheIndex, CacheIndex> { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
this functions need comments inside
I ran the benchmarks locally, and changes are within the noise threshold (not in the sense of criterion, which shows +- 5 to 10% depending on the benchmarks, but that's unfortunately something common, which is also observed across run of the same benchmarks on the same branch). The mantis benchmark seemed to take a small hit, but the numbers are volatile enough so that it doesn't seem to be a huge effect. |
This PR implements part of the
IncCache
interface for incremental merging and modifies some functions related to merging records.